Simulations of estimating causal effects using undersmoothed HAL
Seraphina Shi
2022-09-06
Simulation #1
Data structure: \(O = (W, A, Z, Y)\)
- U - exogenous variables
- W - baseline covariate that is a measure of body condition
- A - treatment level based on W, continuous between 0 and 5
- Z - intermediate curve based on W and A
- Y - outcome, indicator of an event ?
Underlying data generating process, \(P_{U,X}\)
- Exogenous variables:
- \(U_A \sim Normal(\mu=0, \sigma^2 =
1^2)\)
- \(U_A \sim Normal(\mu=0, \sigma^2 =
2^2)\)
- \(U_Z \sim Uniform(min = 0, max =
1)\)
- \(U_Y \sim Uniform(min = 0, max = 1)\)
- \(U_A \sim Normal(\mu=0, \sigma^2 =
1^2)\)
- Structural equations F and endogenous variables:
- \(W = U_W\)
- \(A = bound(2 - 0.5W + U_A, min=0,
max=5)\)
- \(Z = \mathbf{I}[U_Z < expit(2-W-A)]\)
- \(Y = \mathbf{I}[U_Y < expit(W + 5*A + Z - 0.5 * W * A - 8)]\)
- \(W = U_W\)
## W A Z Y
## Min. :-3.621518 Min. :0.0000 Min. :0.0000 Min. :0.0000
## 1st Qu.:-0.679262 1st Qu.:0.6206 1st Qu.:0.0000 1st Qu.:0.0000
## Median : 0.009607 Median :2.0124 Median :0.0000 Median :1.0000
## Mean : 0.000817 Mean :2.1258 Mean :0.4865 Mean :0.5984
## 3rd Qu.: 0.669635 3rd Qu.:3.4071 3rd Qu.:1.0000 3rd Qu.:1.0000
## Max. : 3.992957 Max. :5.0000 Max. :1.0000 Max. :1.0000## Summary of A given W < -1:## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.000 1.359 2.780 2.723 4.215 5.000## Summary of A given -1 < W <= 0:## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0000 0.8095 2.2315 2.2738 3.5758 5.0000## Summary of A given 0 < W <= 1:## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0000 0.4891 1.8463 1.9809 3.1937 5.0000## Summary of A given 1 < W:## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.000 0.000 1.160 1.531 2.591 5.000
##
## Call:
## glm(formula = Y ~ W + A + W * A + Z, family = binomial, data = obs)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -3.8257 -0.0711 0.0006 0.0541 3.6169
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -8.0291 0.2560 -31.361 < 2e-16 ***
## W 1.1818 0.1871 6.317 2.67e-10 ***
## A 5.0013 0.1459 34.269 < 2e-16 ***
## Z 1.0548 0.1193 8.838 < 2e-16 ***
## W:A -0.5794 0.1234 -4.697 2.64e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 13473.1 on 9999 degrees of freedom
## Residual deviance: 2543.3 on 9995 degrees of freedom
## AIC: 2553.3
##
## Number of Fisher Scoring iterations: 9## [1] " MSE: 379.9254, AUC: 0.9898"
n = 200
CV HAL
results
## CV selected lambda (from one sample): 0.0086251603748726
1000 repetition
## The average of CV selected lambdas (from 1000 sample): 0.00793684388619338 The average of CV selected lambdas (from 1000 sample): 0.00792057260113451 The average of CV selected lambdas (from 1000 sample): 0.00792340491044921## z=1:
## z=0:
Globally Undersmoothed HAL
results
## Undersmoothed lambda: 0.00010990826113378
## which is 0.0127427498570313 * lambda_CV
1000 repetition
## The average of unsersmoothed lambda (from 1000 sample): 0.000110282693674584
## which is 0.0144431693912157 * the average of 1000 lambda_CV## z=1:
## z=0:
Oevr a grid of lambda scalers
## TableGrob (7 x 4) "arrange": 7 grobs
## z cells name grob
## 1 1 (2-3,2-3) arrange gtable[layout]
## 2 2 (4-5,1-2) arrange gtable[layout]
## 3 3 (4-5,3-4) arrange gtable[layout]
## 4 4 (6-7,1-2) arrange gtable[layout]
## 5 5 (6-7,3-4) arrange gtable[layout]
## 6 6 (3-3,4-4) arrange gtable[guide-box]
## 7 7 (1-1,1-4) arrange text[GRID.text.1916]
## TableGrob (7 x 4) "arrange": 7 grobs
## z cells name grob
## 1 1 (2-3,2-3) arrange gtable[layout]
## 2 2 (4-5,1-2) arrange gtable[layout]
## 3 3 (4-5,3-4) arrange gtable[layout]
## 4 4 (6-7,1-2) arrange gtable[layout]
## 5 5 (6-7,3-4) arrange gtable[layout]
## 6 6 (3-3,4-4) arrange gtable[guide-box]
## 7 7 (1-1,1-4) arrange text[GRID.text.2192]
## TableGrob (7 x 4) "arrange": 7 grobs
## z cells name grob
## 1 1 (2-3,2-3) arrange gtable[layout]
## 2 2 (4-5,1-2) arrange gtable[layout]
## 3 3 (4-5,3-4) arrange gtable[layout]
## 4 4 (6-7,1-2) arrange gtable[layout]
## 5 5 (6-7,3-4) arrange gtable[layout]
## 6 6 (3-3,4-4) arrange gtable[guide-box]
## 7 7 (1-1,1-4) arrange text[GRID.text.2468]